The broad aim of this five year study is to advance our understanding of language dysfunction in schizophrenia spectrum disorder using event-related potential, event-related gamma, experimental neuropsychological and clinical measures in three DSM-IV diagnosed clinical groups: chronic schizophrenia (CS), first episode schizophrenia (FES) and schizotypal personality disorder (SPD), and their matched normal controls. The use of the three clinical groups will afford studying both differences and similarities in language impairment across the schizophrenia spectrum. Similarities and differences in language processing in males and females in these three clinical groups will be studied as a secondary goal. In order to arrive at a comprehensive model of language dysfunction in the schizophrenia spectrum, we will use a heuristic model of language in which semantic organization includes three distinct but interacting components referred to as properties, processes, and content of semantic networks. Properties refer to connectivity weights among words, and to the size and the coherence of networks. Processes refer to activation, inhibition, and context use (both local and global). Network contents including words, numbers, and symbols will not be measured. We propose a gradual breakdown in language function with most impairment in CS>FES>SPD. We will test these hypotheses in a five-year study of three schizophrenia spectrum groups: 1) 30 patients with CS; 2) 30 patients with FES; and 3) 30 individuals with SPD, and 30 normal controls matched to each clinical group (n=90). Each control group will be matched for age, gender and parental SES to its respective clinical group and all subjects will be right handed, with English as their primary language. We will use ERP, event-related gamma, experimental neuropsychological, and clinicaldata (Thought Disorder Index TDI), to study language disorder in the three clinical groups. We thus predict over-activation in semantic networks as indexed by the N400, word-pair study at short stimulus onset asynchrony (SOA) in SPD, FES, and CS. The evidence of abnorma inhibitory/context utilization processes will be found in the N400, sentence and paragraph studies, with most severe impairment in CS>FES>SPD. The evidence of abnormal properties in semantic networks will be found in the neuropsychological study of word recall, in the N400, word-pair, short SOA study, and in the event-related gamma data, it FES and CS, but not in SPD. Correlational analyses among the four data sets will seek to characterize the relationships among the four domains of analysis.We predict that females will show less impairment than males on all these measures.